This paper presents new causal estimates of incarceration’s effect on voting, using administrative data on criminal sentencing and voter turnout. I use the random case assignment process of a major county court system as a source of exogenous variation in the sentencing of misdemeanor cases. Focusing on misdemeanor defendants allows for generalization to a large pool of people, as such cases are extremely common. Among first-time misdemeanor defendants, I find evidence that receiving a short jail sentence decreases voting in the next election by several percentage points. Results differ starkly by race: White defendants show no demobilization, while Black and Latino defendants are substantially demobilized. Black defendants show a turnout decrease of about 13 percentage points due to jail time. Evidence from pre-arrest voter histories suggest that this difference could be due to racial differences in who is arrested. These results paint a picture of large-scale, racially-disparate voter demobilization in the wake of incarceration.